Regional vs. global temperature calibrations for lacustrine BrGDGTs in the North American (sub)tropics: Implications for their application in paleotemperature reconstructions

Abstract

Branched glycerol dialkyl glycerol tetraethers (brGDGTs) have shown great promise in lacustrine temperature reconstructions across different continents. While brGDGTs have been reported from many different regions and global brGDGT-temperature calibrations have been developed with various methods, southern North America remains an understudied area with little available data. In this study, we analyzed 101 lake surface sediment samples across Mexico and Central America and compared their distributions with those in other lacustrine systems. Nine major brGDGTs were found in all samples. We investigated the relationships between the distribution of the fractional abundances of the nine major brGDGTs and temperature and developed regional calibrations for Mean Annual Temperature using three different approaches, including a novel machine learning method – Ridge Regression. All the regional calibrations provide similar results with very close error ranges (RMSE = 3.1 °C). The majority of global brGDGT-temperature calibrations tend to reconstruct lower temperatures when it is below 15 °C. Interestingly, regional brGDGT calibrations appear to reduce the “cold bias”, but the various global and regional calibrations tested here are not significantly different in their predictive capability.